Skip to content

Bhatt DL, et al. A controlled trial of renal denervation for resistant hypertension. N Engl J Med. 2014;370:1393-401. PMID 24678939

One-paragraph summary

535 patients with severe resistant hypertension — office systolic ≥160 mm Hg and 24-hour ambulatory systolic ≥135 mm Hg on maximally tolerated doses of at least three drugs including a diuretic — were randomised 2:1 to catheter-based renal artery denervation with a single-electrode radiofrequency catheter or to a sham procedure consisting of renal angiography alone. At six months, office systolic pressure fell 14.13±23.93 mm Hg after denervation and 11.74±25.94 mm Hg after sham, a difference of −2.39 mm Hg (95% CI −6.89 to 2.12; p=0.26 against a 5 mm Hg superiority margin). The 24-hour ambulatory systolic difference was −1.96 mm Hg (−4.97 to 1.06). Safety did not differ between groups. Prior uncontrolled studies had reported reductions of 25–30 mm Hg.

Key findings

  • No significant benefit of denervation over sham on either the primary office endpoint or the ambulatory secondary endpoint.
  • The sham arm fell 11.74 mm Hg — approximately the size of the entire effect previously attributed to the procedure in unblinded studies.
  • Safety was acceptable; the trial was negative on efficacy, not on harm.
  • Post-hoc analyses identified predictors of response (PMID 25400162), race-stratified results (PMID 26362830), sleep-apnoea subgroups (PMID 27118620) and 12-month office data (PMID 25835443), none of which rescued the primary result.
  • The 36-month final report (PMID 36130612), after unblinding and permitted crossover, showed larger differences favouring denervation, but rests on imputation for crossovers rather than on preserved randomisation.

Limitations

  • The single-electrode catheter and non-standardised ablation technique were superseded; later trials used multi-electrode radiofrequency or ultrasound with defined protocols.
  • Medication regimens were not standardised and adherence was not verified, so between-group drug changes could obscure a real effect.
  • Office systolic pressure in severe resistant hypertension is noisy and prone to regression to the mean — a poor primary endpoint, and later trials moved to ambulatory endpoints.
  • Many operators performed few procedures; procedural quality was heterogeneous.
  • A neutral result in this population and with this technique does not exclude benefit in a different population with a different technique — which is precisely what subsequent trials showed.

Why it matters

This is the clearest modern demonstration that an uncontrolled procedural signal of 25–30 mm Hg can be entirely accounted for by the sham response plus regression to the mean. It halted worldwide adoption of a procedure that had already been commercialised, and it is the reference case for requiring sham control in device cardiology. It is equally a case against over-generalising from one negative trial: redesigned, sham-controlled second-generation trials subsequently found a real but modest effect of about 4–6 mm Hg ambulatory systolic (PMID 32234534; PMID 36853627). Both errors — over-claiming and premature abandonment — are visible in the same story.

Cited by wiki pages

  • overview
  • device and interventional therapy